I have an idea to automatically kill and launch puppeteer every 20 minutes. Puppeteer browser (chromium) will automatically closed when scrape/parse job finished, and then after 20 minutes, it will relaunch fresh new browser, so it wont eat more resources. I'm using browser pool (cluster) for this method.

import cluster from 'cluster'
import * as os from 'os'
import * as process from 'process'
import * as puppeteer from 'puppeteer'

const launchOptions = {
    args: [
        '--no-sandbox',
        '--disable-setuid-sandbox',
        '--disable-dev-shm-usage',
        '--disable-accelerated-2d-canvas',
        '--no-first-run',
        '--no-zygote',
        '--single-process',
        '--disable-gpu'
    ],
    headless: true
}

let lastProcessUptime = []
let userAgentString = 'Mozilla/5.0 (Macintosh; Intel Mac OS X 10_14_0) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/84.0.4147.125 Safari/537.36'
let viewportSize = { width: 1920, height: 1080 }
let CPUsNum = os.cpus().length
CPUsNum = 1 // Set this number as how many browser page do you want to launch simultaneously

const runThis = async (processID) => {
    const browser = await puppeteer.launch(launchOptions)
    const page = await browser.newPage()
    await page.setUserAgent(userAgentString)
    await page.setViewport(viewportSize)
    await page.setRequestInterception(true)
    page.on('request', (request) => {
        let blockedResources = ['stylesheet', 'font', 'image']
        if (blockedResources.includes(request.resourceType())) {
            await request.abort()
        } else {
            await request.continue()
        }
    });
    await page.goto(url, { waitUntil: 'networkidle0' })
    // ... and do stuff
    // ... then close the page and browser and also kill the process
    await page.close()
    await browser.close()
    lastProcessUptime.push({
        pid: process.pid,
        uptime: Math.round(process.uptime() * 1000)
    })
    process.kill(processID)
    process.exit(0)
}

if (!cluster.isWorker) {
    console.log(`Primary ${process.pid} is running`)

    // Fork workers.
    for (let i = 0; i < CPUsNum; i++) {
        cluster.fork()
    }

    cluster.on('exit', (worker, code, signal) => {
        console.log(`worker ${worker.process.pid} died`) /*with code: ${code} and signal ${signal}*/
        setTimeout(() => cluster.fork(), (20 * 60 * 1000) - lastProcessUptime[lastProcessUptime.findIndex(process => process.pid === worker.process.pid)].uptime) //lets recreate the worker that died
    })
} else {
    let processID = process.pid
    console.log(`Worker ${processID} started`)
    ;(async (processID: number) => {
        await runThis(processID)
    })(processID)
}